Anda di halaman 1dari 10

Design and Construction Aspects of a Large Metro Station Cavern in

Urban Environment.
Plan et construction d’une caverne souterraine d’une station de métro au sein de l’espace

A.Alexandris , P.Vettas OTM Consultants, Athens, Greece.

A.Aranitis, Pantechniki S.A., S. Notarianni Impregilo S.p.A.
K.Boronkay, Attiko Metro S.A.

A large underground cavern, 21 m wide and 110 m long, with 9 meters of cover, has been mined to house an
underground metro station for the Athens (Greece) Metro network. The station has been designed and con-
structed within one of the contracts for the extension of line 3 towards the western suburbs of the city. The
large underground opening was mined sequentially following a ‘central drift procedure’ which proved to be a
very efficient procedure in terms of settlement control as well as in terms of construction time and cost. The
purpose of the present paper is to describe the design and construction process as well as to present the actual
performance of the structure.
Une large caverne souterraine, large de 21 mètres et longue de 110 mètres, et de profondeur de 9 mètres, a été
creusée pour accueillir une station du métro d’Athènes en Grèce. Cette station a été conçue et construite dans
le cadre d’un des contrats pour l’extension de la ligne 3 vers la banlieue ouest d’Athènes. La caverne souter-
raine a été creusée progressivement selon le procédé de la ‘galerie centrale’ qui s’est montré être un procédé
très efficace pour limiter les tassements` ce procédé a été également efficace sur le plan des délais et du bud-
get prévu. Cette communication présente le plan et la construction ainsi que la performance de cette méthode.

Keywords: Underground Cavern, Metro station, Ground settlements.

1 INTRODUCTION Commercial and residential buildings situated in

the vicinity of the station are exposed to settlement
The “Egaleo” station is the last station of a construc- damage risks induced by the tunneling works. Strict
tion contract for the extension of line 3 of Athens limits to surface ground settlements and angular dis-
(Greece) Metro system towards the western suburbs tortions have been set to control those risks.
of the city. For functional reasons (track alignment) A general layout of the station complex is pre-
this station has been designed with a central plat- sented in figure 1 and a cross section of the station
form and an increased track to track distance of 15.0 cavern is presented in figure 2.
m. This requirement leads to the construction of a
very wide underground cavern with a net excavation
span of 21 meters. The large span and the small
overburden (9 to 12 meters) made the design and
construction of the cavern particularly challenging.
The underground part of the station is located di-
rectly underneath a main street of the suburb, which
carries heavy traffic required to remain uninter-
rupted by the construction works. For that reason al-
though a cut-and-cover method of construction was
feasible due to the low overburden, an underground
Figure 1 General layout of the station complex. Shaded areas
mining method was selected for the construction. represent buildings.
Figure 3 Polished slickensided surfaces of the meta-siltstone
unit of the Athens Schist formation

2.2 Geotechnical assumptions and design

Figure 2 Cross section of the Egaleo station cavern parameters
In the vicinity of the station cavern, the Athens
schist formation is below a 3-4 meter thick layer of
2 LOCAL GEOLOGY AND GEOTECHNICAL recent fill. The quality of the rock mass was not
CONDITIONS possible to determine in advance for the entire
length of the station cavern, solely on the basis of a
small number of exploratory boreholes, given the
2.1 Local geology natural heterogeneity of the geologic formation and
its tendency to change abruptly in small distances
The underground part of the station is located within
due to the presence of structural features of tectonic
the Athens Schist formation, which covers most of
origin (faults e.t.c.). For this reason three design
the area of the Athens basin and has been encoun-
ground profiles were foreseen, namely a good (rock
tered during most of the tunneling works for the pre-
C), medium (rock D) and poor (rock E) quality rock
viously constructed part of the Athens Metro (Kav-
mass, covering the extremes of the expected ground
vadas et al. 1996, 1999). Athens schist, according to
conditions. Since the tunneling procedure could not
Koukis & Sabatakakis (1999), is a sequence of upper
adjust during construction to the ground conditions
cretaceous flysch-type meta-sediments, which have
encountered at tunnel face, due to the staged excava-
undergone low grade metamorphism. The basic units
tion of the cavern, the design had to be conservative
of the Athens Schist are meta-sandstones interbeded
enough, to cope with the most adverse scenario.
with meta-siltstones, and in some localities black
However the more optimistic scenarios were also
shales and limestones. In places ophiolitic bodies are
studied in order to bracket the expected performance
also found within this formation. The Athens Schist
of the tunneling procedure.
formation has been subjected to intense folding and
The GSI system (Hoek et al., 1998) was used as a
thrusting during the Eocene and subsequently sub-
tool to characterize the rock mass during tunneling
jected to extensive faulting and fracturing. This tec-
and the Hoek & Brown failure criterion was used to
tonic procedure is responsible for the very complex
derive strength and deformability parameters for
geologic structure encountered during tunneling
each rock mass class on the basis of the procedure
established by Hoek & Brown (1997). The design
From the engineering point of view, the forma-
values for the three rock mass classes were derived
tion presents frequent changes of lithological facies
by Kavvadas (2003) and are summarized in table 1.
at sort distances (at the scale of engineering struc-
tures) as well as an irregular alteration and weather- Table 1. Rock mass classes and respective design parameters
ing pattern. Consequently, the character of the indi-
vidual facies in terms of strength varies from hard GSI E ν c φ
rock to stiff soil. MPa kPa °
Class C 25-35 750 0.30 110 35
Class D 20-25 500 0.30 90 31
Class E 15-20 300 0.30 70 29
3 CONSTRUCTION METHOD AND reconnaissance of the geologic conditions along the
SEQUENCE entire length of the underground structure and gives
an indication of the settlements that the tunneling
works might produce. Of course significant devia-
3.1 Design and construction issues tions from predicted settlement values at this stage
In variable and strongly heterogeneous poor quality can lead to additional support measures and/or modi-
rock masses, like the Athens Schist, the division of fications of the tunneling and support method.
the cavern section to smaller headings is essential
for the effective control of face stability. Block falls
and wedge sliding along slikesides, joints, or weak
zones, is a constant hazard during tunneling in Ath-
ens schist. Ground movements are also controlled
The analysis of the tunneling procedure has been
better reducing the area of the advancing face, since
undertaken by means of two and three dimensional
strains occurring ahead of the face contribute in
finite difference models, using the code FLAC, and
many cases significantly to the final surface settle-
FLAC 3D. The Mohr-Coulomb constitutive model
was used for the rock material. The placement of
For very large sections (in our case the total face
shotcrete in layers as well as its progressive harden-
area of the station cavern is 254 m2) the tunneling of
ing was taken into account in the calculations by
the cavern section in stages is necessary to reduce
changing the respective properties at each construc-
the quantities of excavated material, reinforcement
tion stage. After a series of preliminary analyses the
and shotcrete placing, per round. Limiting excava-
shotcrete shell was decided to be 40 cm thick and re-
tion mucking and support time allows earlier support
inforced with lattice girders and steel wire mesh. A
and reduces face instability risks especially when the
pattern of passive rock bolts was also incorporated
stand-up time is small.
in the design and considered in the analyses.
In the first place it was decided to excavate and
support the cavern from top to bottom and its section
was subdivided in top heading, bench and invert. 4.1 Two dimensional analyses
Moreover, the sequential excavation of the top head-
ing was also deemed necessary considering its large A basic set of two dimensional analyses for the
span and the low overburden of the cavern. The pro- three rock mass classes (representing three distinct
cedure which was employed for the excavation of scenarios) was performed. The calculated maximum
the top heading, foresees the excavation of a central surface settlements were within acceptable limits
drift with vertical side walls reinforced with heavy reaching 11.0mm, 16.0 mm and 30.0 mm for rock
steel beams (HEB sections). Consequently the side mass classes C, D and E respectively. The angular
drifts are mined, and the partition walls are retained distortions were also well controlled by the excava-
acting as columns and stiffeners of the crown vault. tion procedure adopted. Interestingly enough, it was
When all the drifts are excavated, the partition walls observed that an abrupt jump to the evolution of the
or pillars are gradually removed. During this critical surface settlements occurs when the partitioning
stage a series of full height “windows” are opened walls (or pillars) are removed. This action reduces
on the partition walls and a final layer of steel mesh strongly the stiffness of the support system and leads
reinforced shotcrete is applied along the entire directly to settlements. However the fact that the re-
length of the crown area. This continuous layer of moval of the pillars takes place in a sequential man-
shotcrete reduces the risk of defective connections of ner and under controlled conditions (full and easy
the shotcrete shell. In the final stage of the top head- access to construction equipment at this construction
ing construction the remaining pillars are removed stage) it was considered as an advantage of the
and a final layer of mesh reinforced shotcrete is ap- method. In reality creep and relaxation of the shot-
plied at the remaining zones. The procedure (also crete shell and the surrounding rock mass allows
presented in figure 4) is well suited only for rock some redistribution of stresses before the removal of
like formations where the tunnelling of a central the pillars making the effect less prominent.
drift with vertical sidewalls is feasible. Large bend-
ing moments in the vertical sidewalls might develop
in a soft material making this method less attractive. Table 2. Mean and standard deviation of rock mass properties
The excavation of a central drift, along the entire considered in the analyses.
length of the station, provides the opportunity to E ν c φ
drain the surrounding rock mass and minimize the MPa kPa °
problems associated with water inflow, as well as to Mean value 500 0.30 90 31
pre-reinforce the surrounding rock mass with fiber- Standard Deviation ±100 0 ±10 ±1
glass nails. It acts also as a pilot drift permitting a
Figure 4 Top heading construction sequence of the station cavern. Notice the opening of ‘windows’ in the third construction stage
and the gradual removal of the pillars.

As mentioned in paragraph 2, one of the domi- zone follows, as expected, the weaker zones. The
nant characteristics of Athens schist is the abrupt more vertical is the pattern, the easiest it is for the
change of rock mass quality due to lithological plastic zones to reach the ground surface with a di-
variations, structural features and weathered zones. rect impact to surface settlements and mainly to an-
Localized weak zones may have a strong effect on gular distortions. Settlement troughs and angular dis-
shotcrete shell distress and surface settlements. In tortions are compared in figure 6 where reference
order to investigate their effect a second set of two curves for class D and class E are also shown for
dimensional finite difference analyses with ran- comparison. The sub-vertical pattern, which is the
domly varied rock mass properties was performed. worst case, lead to settlements not higher than rock
In these models strength and deformability proper- class E (worst anticipated scenario), while angular
ties were varied according to a Gaussian (Normal) distortions are increased in comparison with those
distribution. In order to produce grid independent derived by the assumption of homogeneous strata.
patterns consistent with the geologic structure of the
Athens schist, the properties were varied in a part 4.2 Three dimensional analyses
random part periodic fashion. A sub-horizontal pat- A two dimensional analysis does not model ex-
tern an inclined pattern and a sub-vertical pattern, plicitly the response of the ground ahead of the ad-
presented in figure 5, were chosen as representative vancing heading, neither does it give any indication
of the possible structural forms of the geologic for- on ground yielding ahead of the tunnel face. Two
mation. The strength and deformability properties dimensional analyses neglect the contribution of set-
used in those analyses are summarized in table 2 tlements due to face compliance and account for
(mean values are those of class D, GSI=20-25). them indirectly through the first deconfinement
The results of the three different random patterns stages. A three dimensional model was set up and
are presented in figures 5. Shear strains and plastic analyzed with FLAC 3D, for the employed method
of staged construction, in order to check the validity
of the two dimensional analyses on which the actual
design has been based. The finite difference mesh
and snapshots of the construction sequence consid-
ered are reproduced in figure 7.
The resulting surface settlements showed a re-
markable agreement with those derived by the two
dimensional analysis for the same ground strength
and deformability parameters. The evolution of the
vertical displacements calculated by the two and
three dimensional analyses for class E (the most ad- Figure 5 Variation of rock mass properties (upper row) accu-
verse scenario) are compared in figure 8. Shotcrete mulated shear strains (middle row) and plasticity indicators
shell axial forces and bending moments were also in (lower row) for the three patterns examined.
general agreement. The pattern of the vertical dis-
placements induced by tunneling and calculated by
the three dimensional analysis is presented in fig. 9.

Figure 6. Vertical Displacements and Angular Distortions at the ground surface for the three variability patterns examined. Settle-
ment troughs for classes D and E are also shown for comparison
Figure 7 Three dimensional model of the station cavern
Construction Stage

The good agreement of the two models (2D and 0

0 1 2 3 4 5 6 7

3D) can be attributed to the fact that for the given set
of strength parameters (even in the worst case sce-

entsat tunnel axis(m

nario), ground yielding does not play the dominant 10

role in terms of surface settlements and the induced
displacements depend mainly on the ground material 15

and shotcrete shell stiffness. In such cases with a


reasonable selection of the deconfinement ratios for Flac_3d

each support stage, a two dimensional model is 25


likely to lead to satisfactory results. Of course if 30

softer ground material is present, a three dimen-

sional analysis might be necessary to investigate the Figure 8 Comparison of the evolution of maximum surface set-
tlement calculated by means of the two and three dimensional
effect of soil yielding ahead of the advancing face. model (Rock E - GSI 15-20)
Beginning the excavation of the central drift

Tunneling the central drift

Tunneling the side drifts

Opening of “windows” and application of shotcrete

Complete removal of pillars

Figure 9. Pattern of vertical displacements derived by a three
Figure 10. Tunnelling the station cavern in stages.
dimensional analysis. (Rock mass class C - GSI 25-35)
5 CONSTRUCTION AND PERFORMANCE During construction, the continuously monitored
surface settlements were compared with the pre-
The construction proceeded without significant dicted figures of the numerical analyses (2D or 3D),
problems and surprises and it was found that the de- in order to assess the safety of the underground
sign was detailed in such a way to avoid construc- opening and of the buildings located at ground sur-
tability problems. The progressive construction of face.
the station cavern is presented in the photographs of In figure 14 the results of the three dimensional
figure 11. analysis for the most relevant numerical model (rock
The adopted tunneling procedure, allowed high class C), and for four characteristic construction
advance rates and permitted the completion of the stages, are presented (dashed lines) and are com-
tunneling works of the station ahead of schedule. pared with the average settlement troughs induced
The tunneling started from the access chamber by the actual tunneling works. The comparison
which was located at the middle of the cavern (see shows that the numerical model provided reasonable
also figure 1) and advanced subsequently towards predictions for the final stages but less satisfactory
both ends of the cavern. The advance rates achieved for the first ones. One possible explanation is that
at each construction stage are presented in figure 11. some dewatering settlements that eventually took
The encountered Athens Schist was composed by place during the first excavation stages and which
meta-sandstone/meta-siltstone alternations, together has not been considered explicitly by the numerical
with lenses or irregular bodies of meta-sandstone analysis, increased the actual settlements observed
(figure 12). Meta-sandstone/meta-siltstone alterna- during the first construction stages. Solid lines in
tions exhibit a well developed, anastomosing folia- figure 14 represent corrected settlements where a
tion which often appears slickensided. Foliation Gaussian curve representing the dewatering settle-
generally dips with low angles towards the east, yet ments has been added to the calculated curves.
its direction varies strongly in short distances due to
extensive folding of the whole formation. Numerous
small to large scale, low angle shear zones and high 6 CONCLUSIONS
angle fault zones crosscut the whole formation. Fault
gouge and cataclastic zones of a few centimeters The central drift method proved to be a very effi-
thick generally mark the shear and fault zones. cient method of a large cavern staged excavation in
The blocks of meta-sandstone exhibit poorly urban environment, where settlement control and re-
developed foliation and are characterized as mas- liability were of outmost importance, while con-
sive. Closely to moderately spaced joints character- struction time and cost had also to be optimized. It is
ize the structure and deformation of the meta- recognized that the application of the particular
sandstone blocks. The GSI values ranges from 25 to method is limited to rock-like formations and it
35 for the meta-sandstone/meta-siltstone alterna- should be avoided if more plastic materials are pre-
tions, whereas for the more competent meta- sent. However it suited very well to the actual condi-
sandstone blocks, the GSI values ranges from 30 to tions of the particular project leading to a very effi-
45. cient construction. Numerical analysis proved to be
In terms of ground water conditions it was a valuable tool to verify the procedure and to dimen-
found that the broader area of the project is charac- sion the support system while during construction
terized by low capacity perched aquifers, developed provided a basis to assess the safety and perform-
almost entirely in the more permeable meta- ance of the tunneling works.
sandstone bodies, which were drained soon after the
tunneling of the central gallery. The water was flow-
ing through fractures (faults, shear zones, joints) ACKNOWLEDGEMENTS
within the meta-sandstone bodies and groundwater
conditions within the tunnel were characterized as The first Author wishes to acknowledge the help
‘dripping’ or ‘dump’ and rarely ‘wet’. of his colleagues in OTM consultants who partici-
During excavation at the various stages, the pated in the project. Prof. M.Kavvadas from NTUA
tunnel face was generally stable and no face buttress provided design values for the Athens Schist. Mr
was needed. A few small-scale over breaks (max A.Alivisatos and Mr A.Mesbouris kindly provided
1m3 of volume) were recorded and were all structur- the monitoring data. The authors would like to ex-
ally controlled. press their gratitude to “Attiko Metro S.A.” for their
The surface settlements were very effectively permission to publish the results of this work. Opin-
controlled by the support measures foreseen by the ions presented herewith do not necessarily reflect
design. Surface settlement contours after the exca- those of the owner “Attiko Metro S.A.” or the con-
vation of the top heading and after excavation of the tractor “J/V-AKTOR S.A.-Impregilo SpA”
complete section are presented in figure 13.
Figure 11 Achieved rates of progress for each section of the cavern.

Figure 12. Encountered geology along the station cavern. A summary of convergence measurements (tunnel crown vertical dis-
placement) and maximum values of surface settlements are also shown.
Top Heading

Bench and Invert

Figure 13 Settlement contours after excavation of the top head-

ing (upper figure) and after the excavation of the complete sec-
tion (lower figure).


Alexandris A.P., M.T.Amparioti & A.N.Pavlou (2006) Nu-

merical Analysis of a Large Underground Cavern Con-
structed with the Central Drift Method. 4th Flac Symposium
Madrid Spain.
Hoek E. and Brown E.T. (1997) practical estimates of rock
mass strength. Int.J. of rock Mech. And Mining Science &
Geomechanics Abstracts, vol 34 No 8 pp 1165-1186.
Hoek E., P.Marinos, M.Benissi (1998) Applicability of the
Geological Strength Index (GSI) Classification for Very
Weak and Sheared Rock Masses-The case of the Athens
schist formation. Bull. Eng Geol. Environ 57:151-160.
Kavvadas M. (1999) "Experiences from the Construction of the
Athens Metro Project", Proc. 12th European Conference of
Soil Mechanics and Geotechnical Engineering, Amsterdam,
June 1999, Invited lecture, Vol 3, pp 1665-1676.
Kavvadas M., 2003. Geotechnical Interpretation Report for the
Egaleo Station Complex. Report submitted to Attiko Metro
Kavvadas M., L.Hewison, P.Laskaratos, C.Seferoglou and
I.Michalis (1996). Experiences from the construction of the
Athens Metro, Proc. Intern.Symp. on the Geotechnical As-
pects of Underground construction in Soft Ground, Mair &
Figure 14 A comparison of the calculated settlement troughs
Taylor (eds) Balkema, Rotterdam.
(FLAC 3D-Rock class C) with the observed settlements.
Koukis G. & N.Sabatakakis (1999) Engineering Geological
Dashed lines are numerical predictions which neglect the effect
Environment of Athens Greece. Bull.Eng.Geol.Env.
of dewatering. Solid lines account for some dewatering settle-
ments calculated from the excess settlements of the first con-
struction stage.